W3C home > Mailing lists > Public > www-international@w3.org > April to June 2008

Re: [Comment on WS-I18N WD]

From: Felix Sasaki <fsasaki@w3.org>
Date: Sat, 14 Jun 2008 20:29:10 +0900
Message-ID: <4853AB86.6070301@w3.org>
To: Dan Chiba <dan.chiba@oracle.com>
CC: www-international@w3.org

Dan Chiba さんは書きました:
> Practically BCP 47 is also a locale identification scheme and using 
> "-" for both #1 locale and #3 language is preferred, for consistency. 
> I think accepting both is a good idea, and more important than which 
> is the standard.

I agree. Addison asked to decide whether we should use "-" (BCP 47 like) 
or "_" (LDML like) as a delimiter. We could choose "-" but make explicit 
that "_" might be used too if people want to be compliant to LDML.

Felix

>
> Regards,
> -Dan
>
> Felix Sasaki wrote:
>>
>> Currently we say in sec. 3.2 about the i18n:locale element
>>
>> Its value MUST be either a valid [LDML] locale identifier or one of 
>> the values "$neutral" or "$default".
>>
>> Dan said about "locale" information in his comment just "already 
>> defined". So I'd like to hear from Dan how important it is for you 
>> that we currently use LDML with "_" or if we could use BCP 47 with 
>> "-", or something else.
>>
>> Felix
>>
>>
>> Frank Ellermann さんは書きました:
>>> Phillips, Addison wrote:
>>>
>>>  
>>>>> For locales names in the language_territory format "_" is
>>>>> AFAIK the standard, compare chapter 8.2 in IEEE Std 1003.1
>>>>>       
>>>  
>>>  
>>>> For POSIX, sure.
>>>>     
>>>
>>> That is what "locale" stands for.  Like "language tag" is what RFC 
>>> 1766 and its successors say, and where we'd use "-".  The
>>> OP wrote:
>>>
>>> | Here is a list of items that we think are common:
>>> |  1. Locale (already defined)
>>> |  2. Timezone (already defined)
>>> |  3. Language (used when UI language is different from the
>>> | language deduced from the UI locale. e.g. "de" for German
>>> | language, "fr-CH" for Switzerland/French locale)
>>> |  4. Collation (based on the IANA collation registry)
>>> [...}
>>>
>>> Maybe he confused the terminology, he needs "language tags"
>>> in (3), and fr-CH is a "language tag".  In point (4) ff. he
>>> mentions some IANA registries, he could also do this in (3).
>>>
>>> But (1) is apparently about locales, not about the language
>>> tags covered in (3).  So in (1) we'd say fr_CH, not fr-CH.
>>>
>>> That is an important difference, locales come with various
>>> settings down to currency symbols, but there are not many
>>> to pick from.  OTOH language tags are only about languages
>>> and maybe scripts, and there are lots of valid no-nonsense
>>> combinations.  
>>>> there are other locale systems where this isn't the case
>>>> or for which the separator is indeterminate. There is *no*
>>>> definition of 'locale' for the Web and/or Internet
>>>>     
>>>
>>> Well, when I look at the CLDR pages they use unsurprisingly
>>> "_", not "-".  That's arguably two standards, POSIX and CLDR.
>>>
>>>  
>>>> There is no particular reason to use POSIX locales on the
>>>> Internet and there is some history of abusing BCP 47 for
>>>> the purpose already.
>>>>     
>>>
>>> Disagree, I see no reason to "abuse" the IANA language subtag 
>>> registry for something it is not, a locale registry, because
>>> there is already a CLDR with different goals.  
>>>> If we allow underscore is may actually be harmful, since it
>>>> may promote the possibly-erroneous assumption that we mean
>>>> POSIX locales.
>>>>     
>>>
>>> Or CLDR locales.  It's a rather useful difference, "i-default" is no 
>>> locale, and "C" is no human language.  With "en_GB" I'd
>>> get an odd (from my POV) date format, with "en_US" I lose the
>>> metric system, get alien temperatures, and a currency backed
>>> by hot air.  Which isn't my plan when I say "en-GB" or "en-US".
>>>
>>>  Frank
>>>
>>>
>>>   
>>
>>
>
>
Received on Saturday, 14 June 2008 17:51:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 19:17:17 GMT